Danaher and Microsoft partner to bring AI control and guidance to lab instruments
Initial applications are expected by the end of 2026. Early microscope results suggest faster work for novice scientists, but the figures come from preliminary internal testing.
Danaher and Microsoft are building an Azure-backed platform for AI agents that can work across lab instruments, data and workflows, with researchers retaining oversight of equipment use and recommendations. The staged roadmap puts selected applications in customers’ hands before the end of 2026 and agent-enabled instruments in 2027. The approach could make lab automation more integrated, but the early performance figures come from internal microscope testing and do not establish results across other instruments or laboratory work.
01
In preliminary internal tests, novice Leica microscope users with the AI assistant completed an experiment up to 83% faster and with up to 68% fewer steps than novices without it.
02
The companies also reported high-quality data and completion times comparable to an expert; the comparison does not establish equivalent performance across other tasks.
03
Molecular Devices’ MetaXpress Acquire AI Assistant is among the applications expected to reach customers in 2026.
Danaher and Microsoft have announced a partnership to connect scientific instruments, experimental data and laboratory work through AI agents—software designed to carry out tasks. The companies want those systems to help scientists plan experiments, operate equipment and interpret results, with initial commercial applications expected to reach customers before the end of 2026.
The agreement centers on a common platform for instrument control and scientific insights. It is intended to connect technologies across Danaher’s businesses while remaining open to the wider laboratory environment. That scope makes connections between instruments, data and workflows part of the project, rather than treating each instrument as an isolated tool.
From scientific intent to the next experiment
Microsoft Azure will provide underlying technology for the ecosystem. Danaher brings its expertise in scientific instruments and laboratory workflows. The planned systems span four stages of experimental work, reaching beyond setting up an instrument to assessing what it produces and proposing what a researcher should try afterward.
Experimental planning: Translate a scientist’s intent into an experimental plan, including instrument setup and methods. This is the proposed starting point for turning a research question into work on laboratory equipment.
Instrument control: Operate hardware within defined safety boundaries and under user oversight. The tools are also intended to guide execution and help troubleshoot problems as researchers carry out an experiment.
Real-time interpretation: Analyze results while an experiment is running and identify potential quality problems. The stated role includes helping scientists recognize issues during the process, rather than only examining results afterward.
Next-step recommendations: Suggest corrective actions and subsequent experiments for scientists to review. These recommendations are meant to help connect what researchers learn from one experiment with decisions about the next.
The deployment design spans both the instrument and the cloud. Luciano Lucas, Danaher’s head of autonomous and intelligent labs, described a common platform that would let the company deploy device-specific agents across its instruments. Some would run at the instrument itself, while others would operate in the cloud.
Early microscope results come from internal tests
The companies’ early efficiency claims concern novice scientists conducting an experiment on a Leica microscope. Healthcare Digital described preliminary internal testing comparing users working with an AI assistant against novices working without it. The reported savings are maximum reductions, not average improvements across laboratory work.
Company-reported microscope experiment
Up to 83%Time reduction
Novices using the AI assistant completed the experiment in up to 83% less time than novices without it.
Up to 68%Fewer steps
The AI-assisted experiment required up to 68% fewer steps in the preliminary internal comparison.
The companies also reported high-quality data and completion times comparable to an expert’s. That expert comparison concerns speed; the stated percentage reductions compare novices with and without assistance. The findings remain preliminary and internally generated, rather than an independent assessment of how the tools perform.
Applications first, new instruments in 2027
Danaher’s rollout separates initial applications from a later generation of equipment. Healthcare Digital identifies Molecular Devices’ MetaXpress Acquire AI Assistant among the applications expected to reach customers in 2026. Next-generation, agent-enabled scientific instruments are planned for 2027; that is a development timetable, not a claim that those instruments are available now.
For scientists, the proposed division of responsibility remains explicit: AI would guide work and recommend actions, while researchers stay involved in decisions. Even the planned recommendations for subsequent experiments are intended for scientists to review.
Reader comments
Newest comments first. Replies stay oldest first.